Gurus,
I need your help to identify whether MINA has memory leak. I am little lost as what is happening. We are currently using MINA 2.0.0-M5. The machine has 16G memory and I allocate 12G for the process. After running for a day under fairly heavy load, it runs into OutOfMemory(top shows that the process takes 75% of memory, which is exactly right). So I took it out of load balancer, but I did NOT take down the process. Three hours later (GC claimed back some memory, it was back to 68% instead of 75%), I took a jmap against the process. It generated more than 10G of dump data. I examined the result and found out that are tons (495,968 in total) of the DefaultWriteRequest (see below) hanging around. Though it is small, the message contained inside is huge. There is no network traffic to the process after cutting it off the load balancer and it has been long enough, any idea why those objects are still hanging around? Your insight is highly appreciated! Yong ------------------------------------------------------------------------ ----------------------------------------------------------------- instance of org.apache.mina.core.write.defaultwriterequ...@0x2aab263fd080 (40 bytes) Class: class org.apache.mina.core.write.DefaultWriteRequest <http://localhost:33333/class/0x2aaaae9afd70> Instance data members: destination (L) : <null> future (L) : org.apache.mina.core.future.defaultwritefut...@0x2aab263ffe60 (61 bytes) <http://localhost:33333/object/0x2aab263ffe60> message (L) : [...@0x2aab263fd0a8 (11690 bytes) <http://localhost:33333/object/0x2aab263fd0a8> References to this object: org.apache.mina.filter.codec.protocolcodecfilter$messagewriterequ...@0x2 aab263fd068 (24 bytes) <http://localhost:33333/object/0x2aab263fd068> : field parentRequest
